Having worked in HIT for more than two decades, I am not an unbiased observer. However, I believe it is not the technology itself which causes clinicians to burnout, but rather the implementation of the technology and the feeling that the mandated EHR documentation effort is doing little to improve the process of care or benefit patients. It is possible for properly designed and implemented EHRs to reduce this frustration and burnout. However, the way information is collected and manipulated in EHRs needs to be radically re-thought.
The challenge of EHR documentation
Back in the 90s, clinicians rarely used EHRs. It was mostly ancillary staff and coders who interacted with the technology since the language employed was a significant barrier to clinician users
. They couldn’t recognize the terminology used by the systems, and the primary use case of billing and reimbursement felt irrelevant to them. This led to the development of clinical interface terminology (CIT) which bridged the gap between the clinician’s intent (in their heads) and the machine. It also made it possible to reduce the amount of manual coding that went into reimbursement and reporting. It didn’t take long for more documentation responsibility to be placed on the shoulders of the clinicians. This requirement was often sold with the promise that the information was needed for patient care, or that it would be used to improve their work. Yet population health, clinical decision support, and automated clinical trial identification never really lived up to the expectations set by the purveyors of technology. Most EHRs were still primarily designed to create a medical-legal document and to record information for billing and/or auditing.
The case for data transformation
A key challenge that HIT designers face is that the information collected for one purpose usually does not immediately serve other uses without transformation. We have reams and reams of encounter data, raw lab results, histories of procedures, and images. Although perhaps machine-readable, this data is not presented in a format that can be processed quickly by clinicians or even intelligent machines. There is a step missing. For example:
- A clinician might record that a patient has “Stage 2 ER-, PR- Adenocarcinoma of the Breast” on her EHR problem list. A registry might want to simply record that she has breast cancer. A clinical decision support rule might care more that she has Stage 2 cancer, or perhaps that the cancer is hormone receptor negative.
- In another example, the same data may be required for several incentive or federally-mandated programs but the data must be entered in a different form or with different codes.
EHRs could enable data to be collected once and reused multiple times if the data is collected properly and then transformed to meet different use cases. Unfortunately, EHR developers have often chosen to build separate interfaces for each program, therefore requiring clinicians to waste time answering seemingly redundant questions. EHRs (and EHR developers) need help in transforming information so that clinicians can benefit from the improved efficiency and enhanced functionality necessary to offset the burden of collecting all the data. The challenge is that transforming the information requires knowledge of both the problem being addressed and the specific qualities of the data that was collected, a combination of subject matter expertise not common in the industry.
When it comes to data, more isn’t (necessarily) better
I don’t believe that HIT vendors set out to provide poor software or software designed to frustrate and burn out clinicians. They will, however, need to overcome significant technical and historical obstacles in order to deliver on the promise made to clinicians. This will be an iterative process. Moreover, new machine learning and artificial intelligence solutions are not going to be able to overcome this challenge on their own either – but that’s a topic for another post. The same is true of interoperability. The increasing flow of information enabled through recent legislation against information blocking will only make the burnout problem worse. More data, mashed together and presented to clinicians, will further overwhelm them unless properly processed and summarized. IMO continues to play an important role in ensuring that clinician intent is properly captured in our health information systems in the form of accurate, granular data that is intuitive and relevant for clinicians and patients. The challenge now is to transform, aggregate and summarize that data to empower clinician- (and patient-) facing applications which fundamentally improve the quality of care and quality of life for everyone. Working together, I am optimistic this can be done.